Low (dc) resistance isn't the issue: you can actually short your avr's pins without damaging them
You've done this?
FWIW... I've accidentally shorted outputs and even inserted the 328P backwards... never hurt the chip (yet).
I don't think you can claim that - appearing to still work is not the same as undamaged - overloading the output transistors causes local overheating and perhaps electromigration on the chip which will first degrade then disable the transistors in that area - the reliability and robustness is impaired before obvious/gross loss of function is seen. Once degraded the risk of future failure without warning is increased. Manufacturers put values in the abs max spec that are just conservative enough (they don't want to be sued for being unfit-for-purpose, yet want to look as good as possible compared to the competition... I would always recommend staying below them by a sensible margin.
Of course, in reality the datasheet specs MUST be adhered to or else the chip WILL get ruined sooner or later.
A more positive spin is obey the abs max constraints so you end up with about strange unexplaned failures in the future!